Stiffness of the rope is much bigger factor. I have soft 10mm that passes through easier than some stiff 9mms.
Read my full reply. I have email from Madrock assuring using a blocking knot will not result in breaking of the device or the rope. Thus it is impossible to fall further than the knot.
What Madrock says they consider as "failure of the system" means is in such situation you jammed knot, not the cam of the device is holding your full weight. With static rope you also have a risk of increased load. Where the warning is crucially important is in canyoning, where people often rappel through waterfalls. If you have jammed device under a waterfall, you drown. It's their choise not to add nuance to the instructions. I disapprove but that's their choise.
Key thing to take away YOU ARE NOT BETTER OFF TYING FRICTION HITCHES ABOVE THE DEVICE. You create risk of two dangerous situations. One your hitch may jam and you need to go hands free to release yourself. Two you're making it very hard to control the speed of your rappel as you need to manipulate the friction hitch and the handle with one hand and it creates temptation to let go of the brake strand.
I try to condense, by tying a knot that can support your weight jammed against a hole that is the size of Safeguards rope channel, you prevent yourself falling past the knot despite Madrock considering the system not working as intended. You might need a prusik and a foot loop to release the device after, but this will never cause a risk of the rope or the device breaking at loads anyone would want to expose themselves to. Put 1 and 1 together. I have the full reply from Madrock posted somewhere over here too if you're interested. I you're climbing on slack hyperstatic materials you're doing something stupid on it's own but your body will break before Safeguard does.